https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125848
--- Comment #5 from Drea Pinski <pinskia at gcc dot gnu.org> --- (In reply to Jakub Jelinek from comment #4) > That is just weird. Yes, the patch changes the lookup rules for the dtors, > so in some cases there will be a placement operator delete call if exception > is thrown from constructor of a placement new, in other cases there won't be > one where there was one before, but in most cases placement operator delete > just does nothing. > I'd really like to see a preprocessed source where it makes any measurable > difference... Once I will try to get you this but it might not be for a week or two (naveen bisected this for me). And yes i agree this seems odd. For x86, this benchmark seems very noisy too so I can't tell from lnt.opensuse.org graph if there was a jump around your commit.
